What does it mean that the first 30 days are critical?

New agents are most fragile in their first month — they need extra attention and support to survive.

Example: You design a new decision agent: 'Before accepting any meeting, ask whether it advances one of my three quarterly priorities.' The first week, you use it four times and it works well — you decline two meetings that would have wasted three hours. The second week, a senior colleague sends an urgent-sounding invite and you override the agent without thinking. By week three, the agent is effectively dead. It never failed because it was wrong. It failed because it was new, and new agents cannot survive neglect.

Try this: Identify one agent you've deployed in the last 30 days — a habit, a decision rule, a review practice, anything you explicitly designed and started running. Write down: (1) How many times you've actually executed it. (2) What situations caused you to skip or override it. (3) Whether it has a scheduled check-in or is relying on your memory to keep it alive. If you don't have a recent agent, that's your answer — your deployment pipeline has an infant mortality problem.